Something fishy with math. consts?

[IainNZ]

See e.g.

• [PkgEval] Tau may have a testing issue on Julia 0.4 (2015-01-30) JuliaMath/Tau.jl#7: ERROR: LoadError: test failed: (a = 6.2831853071795... == 6.283185307179586)
• https://github.com/jlapeyre/LambertW.jl/issues/1: ERROR: LoadError: LoadError: test failed: (0.5671432904097838 == a = 0.5671432904097...)
• [PkgEval] ZChop may have a testing issue on Julia 0.4 (2015-01-30) jlapeyre/ZChop.jl#2: ERROR: LoadError: test failed: (3.141592653589793 == a = 3.1415926535897...)

[IainNZ]

  | | |_| | | | (_| |  |  Version 0.4.0-dev+2997 (2015-01-30 05:35 UTC)
 _/ |\__'_|_|_|\__'_|  |  Commit e41a507 (0 days old master)
|__/                   |  x86_64-unknown-linux-gnu

[IainNZ]

I think this is #9198 related?

[IainNZ]

@simonbyrne can you do a quick check to see if these errors in these package tests are because they were relying on some behaviour they shouldnt?

[simonbyrne]

Yes: now all numbers should only compare equal if they are actually numerically equal, so pi != float(pi). As MathConsts usually represent irrational numbers, they will never be equal to any floating point values.

[IainNZ]

Excellent